Artificial Intelligence in the Workplace
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Meinungsmonitor
Künstliche Intelligenz
Factsheet No. 3 - November 2020
Artificial Intelligence
in the Workplace
How does the German population perceive the influence
of artificial intelligence on the future of the workplace?
The fear of losing one´s job due to increasing automation of processes and the implementation
of artificial intelligence (AI) is considered the overarching concern for employees when they
ponder the effects of digitization. The following survey shows that these representations only
reflect the actual concerns of the respondents to a limited extent. One of the key findings of the
study is that the surveyed citizens perceive little potential for change in their own workplace.
However, those who do expect changes clearly differentiate between various aspects of work.
They expect positive outcomes of AI has potential when it comes to matters involving occupa-
tional health and safety, the necessary skill requirements and the overall workload. However,
the maintenance of social contacts, their respective income or opportunities for co-determina-
tion in the workplace could suffer according of the respondents. The handling of data by and
potential workplace surveillance through AI are two of the biggest concerns. Only very few peo-
ple fear job loss for themselves, people from their private environment as well as mass unem-
ployment.
Background
A popular opinion both in the academic litera- The assumption that digital technologies of AI
ture and the public debate is that digital technol- will bring about changes in the workplace is only
ogies, such as AI, will result in far-reaching and rarely questioned. Time will tell which progno-
drastic change in the workplace (z. B. Frey & ses are accurate. Also, little is known about the
Osborne, 2017; Makridakis, 2017). For years expectations and fears that already exist
there have been concerns, that the constantly among the population with regard to the influ-
increasing automation of processes will eventu- ence of AI on the future of work. A socially re-
ally result in a loss of jobs. At the same time, sponsible design of digital technologies is most
the overall potential of and the improvement likely if all stakeholders can voice their con-
through digital technologies is repeatedly em- cerns and that their perspectives are incorpo-
phasized - not only for employers, but also for rated into the design process. Time and again,
employees (e.g. by increasing safety standards however, media reports can be found in which
or reducing physical strain). fear of "the robot", of losing one's job or of con-
stant surveillance are discussed which rather
In their respective AI strategies, both the Ger- indicate negative sentiments towards emerging
man government and the European Commis- digital technologies. In the third thematic survey
sion highlight the importance of AI for the work- of the Opinion Monitor Artificial Intelligence
place. Their considerations range from a funda- (German project title Meinungsmonitor Kün-
mental change in work profiles and the creation stliche Intelligenz; acronym used from here on:
of new professions to problems of discrimina- MeMo:KI) we deal with the presumed conse-
tion in AI-supported recruitment processes and
questions of data protection for employees.
Ein Projekt von: Gefördert durch:Meinungsmonitor
Künstliche Intelligenz Factsheet No. 3 – November 2020
quences of the introduction of AI in the work-
place. The survey includes 1,001 respondents,
602 of whom are currently employed.
Methodology
Method Online survey
Ausführendes Institut: forsa Politik & Sozialforschung GmbH
Population: German population over 18 years of age who use the Internet at
least occasionally
Sample: Weighted random sample (N=1.001)
Weighting Criteria: Age, gender, and region (by federal state)
Survey Period: 2020, August, 24-28
Further Information: Detailed Methodology Overview for the MeMo:KI project [in German
language].
Only few people fear the loss of their own jobs through AI; jobs
for social minorities do appear to be at risk
Discussions about the effects of digitization and However, respondents suspect that social mi-
AI work-related issues often focus on the secu- norities will be adversely affected by AI in the
rity of workplaces. Robotisation and automation workplace. 54 percent of those surveyed stated
are often seen as a danger to the workplace. that they expect such social groups to experi-
We therefore wanted to explore how likely re- ence further disadvantages from AI. It remains
spondents perceive a loss of employment for to be seen how exactly that conclusion is
themselves, people from their immediate envi- drawn: is it due to the way AI penetrates the
ronment or members of social minorities and workplace or are the answers due to the aware-
disadvantaged groups such as people with a ness of the often precarious (employment) situ-
migration background, disabled or single par- ation of members of social minorities who are
ents. therefore regarded as more at risk by AI? Is the
predicted job loss mainly associated with pro-
Overall, only a few of the respondents expect fessions that largely rely on minority employ-
fundamental changes in the job market. Only ment? Or is AI assumed to discriminate against
around 11 percent of those surveyed assume specific groups? In a previous survey, we
that more new jobs will be created than will be showed that less than 30 percent of people are
lost, while about 52 percent disagree with this aware of the discriminatory potential of AI. In
statement. The respondents see the least risk this specific case, it seems that the awareness
of job loss in their own line of work. Only 6 per- of the vulnerability of specific groups to the fun-
cent consider the loss of their own job through damental changes that AI can cause is quite
AI probable or very probable. 20 percent sus- high. It is also far more pronounced, than the
pect mass unemployment and slightly less be- awareness to the personal risk situation.
lieve that family members, friends and ac-
quaintances could lose their jobs (18 percent).
Project by: Funded by:Meinungsmonitor
Künstliche Intelligenz Factsheet No. 3 – November 2020
Figure 1: Presumed consequences for the labor market in the next five years.
minorities in society being particularly affected
by loss of jobs 54 20 22 5
mass unemployment looming/threatenin 20 34 43 3
family members, friends and acquaintances
losing their jobs 18 32 46 5
more jobs being created than replaced 11 32 52 5
you losing your own job 6 8 75 3
0% 20% 40% 60% 80% 100%
(very) likely undecided (very) unlikely don't know
Annotation: N=1.001, all values in percent.
Question: How high do you estimate the probability that an increased use of artificial intelligence in the next five years results in,
…
(1=very unlikely; 5=very likely)
AI driven changes are primarily expected in data security
and job requirements
As has already been predicted on several occa- the medium term, in particular regarding the de-
sions, the use of AI can lead to a variety of velopment of income (16 %), co-determination
sweeping changes in the workplace. We asked in the workplace (20 %), career opportunities
the employed respondents (n=602) for their as- (21 %) and job satisfaction (21 %). A further, 42
sessment of the influence of AI on various work- percent expect that there will be changes with
ing conditions in their own professional activi- regard to the security of their data. Lastly, 38
ties. To what extent is there an overall aware- percent of employed respondents suspect
ness within the general population of how AI changes in terms of the expertise required at
could change personal workload, tasks and re- their workplace.
sponsibilities?
Thus, it can be summarized that only a small
It clearly shows that for most working condi- part of the respondents is aware of the potential
tions, a relative majority of the German popula- for change in AI. In some areas it is easier for
tion tends to expect little to no change (Figure the interviewees to perceive AI and their own
2). Only a small proportion of the employees working conditions to be related to each other.
surveyed believe that implementing AI into the According to experts, however, it is very likely
workplace will change something for them in that AI will lead to significant changes in various
Project by: Funded by:Meinungsmonitor
Künstliche Intelligenz Factsheet No. 3 – November 2020
areas. Trade unions and other employee repre- If these processes take place without involving
sentatives need to raise awareness among citizens themselves or if they are only able to
their members and constituents for the potential react to processes that are already underway,
changes in AI (beyond potential job losses). this can lead to friction during the technological
Only comprehensive information about and transformation phase and may very well result
awareness of the possible changes through AI in growing dissatisfaction and dwindling self-ef-
opens the way for employees to actively partic- ficacy on the part of employees.
ipate in shaping the transformation processes.
Figure 2: Assessing change within the own workplace through the introduction and implemen-
tation of AI
data security 42% 30% 23% 5%
required expertise 38% 26% 35% 5%
work load 31% 26% 39% 5%
social contacts 29% 22% 46% 4%
physical health 27% 21% 47% 5%
job security 24% 25% 46% 5%
job satisfaction 21% 35% 39% 6%
career opportunities 21% 26% 47% 7%
co-determination 20% 26% 50% 4%
income 16% 31% 49% 5%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
(very) significant change (4&5) undecided (3) no/little change (1&2) weiß nicht
Annotation: N=602 (Employees), all values in percent.
Question: How would you rate this: To what extent will the following work conditions in your own professional life be changed in
the next five years by the use of artificial intelligence?
(1=no change; 5=very significant change)
Opportunities for the relief of employees through AI
Even though the majority of respondents antic- order to get better insights into the expected
ipate surprisingly little change in their workplace changes in the workplace, we asked the partic-
as a result of AI, the question still arises as to ipants whether they tend to rather see chances
the direction of the presumed change. Do the or risks in using AI. Figure 3 shows the answers
respondents believe that AI will have positive or of those who do suspect a change in the corre-
negative effects on their working conditions? In sponding working conditions.
Project by: Funded by:Meinungsmonitor
Künstliche Intelligenz Factsheet No. 3 – November 2020
Figure 3: Assessment of opportunities and risks in working conditions for affected employees
physical health (n=161) 13% 25% 56% 6%
required expertise (n=218) 21% 25% 51% 3%
work load (n=185) 29% 24% 45% 2%
career opportunities (n=124) 36% 26% 35% 4%
data security (n=251) 44% 21% 34% 2%
job satisfaction (n=123) 34% 32% 32% 2%
co-determination (n=119) 49% 22% 26% 4%
income (n=94) 48% 25% 24% 4%
job security (n=146) 54% 24% 20% 3%
social contacts (n=172) 64% 22% 11% 3%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
(rather) risk (1&2) undecided (3) (rather) opportunity (4&5) don't know
Annotation: N=94-251 (Only includes those people who see significant potential for change); all values in percent.
Question: Do you consider the use of artificial intelligence for the following working conditions in your professional activity more
as a risk or as an opportunity?
(1=risk; 5=opportunity)
From the point of view of the interviewees, AI these are more likely to have negative conse-
has a positive influence especially on occupa- quences.
tional safety and health in the workplace. 56
percent expect positive changes. Likewise, the The results once again showcase differentiated
perceived opportunities predominate in terms of assessments of AI and its influence on different
competence requirements (51%) and workload aspects of work. For empirical research, this
(45%). Most of those who suspect changes means that the individual consequences of AI
here do not see a devaluation of their own qual- and their perceptions by citizens must be inten-
ifications or an excessive demand for new sively researched and examined. Especially
tasks, but rather relief and new possibilities for with regard to explanations for the differences
individual development. in perception. For public discussion and political
decision making, the results give rise to a differ-
However, the data also shows that there is an entiated approach to the potential conse-
acute awareness of the risks of AI for individual quences of AI in the workplace. The results can
areas. This is particularly true for social compo- thus provide an impetus for a more in-depth dis-
nents of work, such as social contacts (64 % cussion involving and drawing from public opin-
risk perception) or opportunities for co-determi- ion. If respondents become more aware of the
nation (49 % risk perception). A sense of be- problem in the areas covered by the survey and
longing, camaraderie with colleagues and so- if the perceived risks in the use of AI outweigh
cial interaction seem to be endangered by AI the potential opportunities, this may lead to a
according to the interviewed employees. The stronger response to AI systems. At the same
results further show that if changes are ex- time, exaggerated hopes can lead to disap-
pected in terms of job security and income, pointment.
Project by: Funded by:Meinungsmonitor
Künstliche Intelligenz Factsheet No. 3 – November 2020
Hardly any worries about a potential lack of expertise
Looking more closely at the fears of the em- workload is hardly concerning to most inter-
ployed persons regarding the influences of a viewees (Figure 4). The situation is different
stronger penetration of the workplace by AI, it when it comes to the transparency of personal
becomes apparent, comparable to the potential data and the threat of surveillance. 50 and 40
for change, that in most areas fears exist only percent of the workforce fear negative conse-
in a small number of the interviewees. Job se- quences in this regard.
curity in particular, as well as an increased
Figure 4: Feared consequences of AI
… there will be less transparency about how
personal data is used 50% 23% 21% 6%
… I will be increasingly under surveillance at
my workplace 40% 24% 31% 5%
… my social contacts with clients and
colleagues at work will suffer 28% 21% 45% 6%
… my work will be less valued - appreciated 26% 20% 49% 6%
… I will have less autonomy in the workplace 25% 25% 44% 6%
… my qualifications are no longer sufficient
for the requirements in my work area 11% 19% 64% 5%
… I am overstrained by the use of AI in my
industry 10% 18% 66% 6%
… I will loose my job 10% 16% 68% 5%
0% 20% 40% 60% 80% 100%
stimme (voll und ganz) zu (4&5) teils/teils (3) stimme (überhaupt) nicht zu (1&2) weiß nicht
Annotations: N=602 (employees), all values in percent.
Question: To what extent do you agree or disagree with the following statements regarding the changes in the workplace through
artificial intelligence (AI)? I am afraid that...
(1=do not agree at all; 5= totally agree)
Implementing AI into the workplace will affect used to the autonomy at the workplace). In con-
individual population groups to different de- trast, younger workers tend to look rather care-
grees. In general, the use of AI can lead to sys- lessly at AI in the workplace. This is interesting
tematic problems of preferential treatment and in that the older respondents are the ones who
discrimination (see also Themenumfrage 2 des are likely to have the least or shortest exposure
MeMo:KI). Therefore, it is also relevant to look to AI and the effects of AI in the workplace. At
at the results of the survey not only in the ag- first glance, they seem to be less affected by
gregate (across all respondents), but also bro- potential risks. The respondents who are likely
ken down into individual population groups. to have a longer working life are also those who
have had more contact with other digital tech-
In terms of workload, job requirements or qual- nologies. This can lead to less fear of contact
ifications and job security, the socio-demo- with AI.
graphic groups hardly differ from one another.
At least two key effects can be identified for the Furthermore, we find an effect that can be at-
other working conditions. On the one hand, tributed to the education of the respondents: It
older employees especially seem to be the turns out that there is no difference in fears re-
ones who have fears about various conse- garding their own job security regardless of the
quences (from the transparency of the data level of education the respondents achieved.
Project by: Funded by:Meinungsmonitor
Künstliche Intelligenz Factsheet No. 3 – November 2020
This is curious because low-wage jobs are seen use of AI, or that it will lead to a diminished ap-
as particularly threatened by automation. How- preciation of their own work. Thus, they are
ever, it can be seen that people with lower edu- more likely to perceive negative consequences
cation degrees in particular are afraid that so- with regard to various aspects of work than peo-
cial contacts will be negatively influenced by the ple who have a higher education.
Table 1: Major fears of consequences of AI in the workplace by population group, in proportion
Lower de-
Abitur or hig-
18-35 years 36-50 years 51-65 years gree than
her
I fear, that… (n=169) (n=220) (n=202) Abitur
(n=419)
(n=180)
… there will be less transpar-
ency about how personal data is 40 % 53 % 56 % 50 % 48 %
used
… I will be increasingly under
35 % 42 % 43 % 39 % 41 %
surveillance at my workplace
… my social contacts with cli-
ents and colleagues at work will 21 % 27 % 36 % 25 % 35 %
suffer
… my work will be less valued -
21 % 25 % 30 % 21 % 36 %
appreciated
… I will have less autonomy in
21 % 25 % 32 % 23 % 29 %
the workplace
Annotation: N=602 (employees), all values in percent
Question: To what extent do you agree or disagree with the following statements regarding the changes in the world of work
through artificial intelligence (AI)?
(1=do not agree at all; 5=totally agree); Shown are the percentage values for all those who indicated 4 or 5 on the scale.
Conclusion
The results of the survey give insights into the An interesting result of the study is that our in-
perceptions of the population as to how the use terviewees do not seem to perceive the fear of
of AI can change the workplace - a topic that is job loss or negative developments on the job
quite prominent in public discourse and is fre- market, which are so often portrayed and dis-
quently accompanied by dystopian prognoses. cussed in the media, as strongly as one would
In summary, the data show - at least from a me- assume. This may be due to the time factor,
dium-term perspective - that the majority of the considering we asked the respondents about
workforce perceives little potential for change or the changes they expected within the next five
that the interviewees assume rather few threats years, but it could also indicate that there is a
from AI. certain openness towards new technologies or
at the very least no increased fear among the
So is AI more an opportunity or a risk with re- general population.
gard to the workplace? According to the as-
sessment of the surveyed employees, it is both! How can the core results of our survey be sum-
Risks by AI at the workplace are seen mainly in marized? On one hand, we found that public
the area of data security and in changing com- opinion is less polarized than expected and
petence requirements. In addition, we find dif- mostly positive. Nevertheless, fears should be
ferentiated judgements depending on the re- taken seriously and made visible. In doing so,
spective aspects of work that were questioned.
Project by: Funded by:Meinungsmonitor
Künstliche Intelligenz Factsheet No. 3 – November 2020
they can be met with more detailed explana- professional associations and trade unions or
tions and information so that fears can be over- politics itself can ensure that a problem aware-
come. Justified concerns can also be discussed ness is created in the population and citizens
in order to find sustainable solutions. Moreover, are enabled to participate in the process, can-
the data indicate that few people have any idea not be answered by this study at this point in
of what AI exactly involves and how its technol- time. However, our data show a necessary prel-
ogies can affect the workplace. Against the ude to a differentiated discussion on how this
background of the many forecasts, opinions complex transformation process can be shaped
and evaluations of experts, the current dis- and how different population groups with their
course on AI points to a need for more infor- specific perceptions not only become visible,
mation about the opportunities and risks that but how their respective outlooks, perceptions
the technology presents. This should eventually and needs are incorporated into this process.
reach those potentially affected so that these
groups can actively participate in shaping the
future of the workplace. Whether the media,
To be cited as: Meinungsmonitor Künstliche Intelligenz (November, 2020). Artificial Intelligence in the Workplace. How does the
German population perceive the influence of artificial intelligence on the future of the workplace? Factsheet No. 3 of the Mein-
ungsmonitor Künstliche Intelligenz [Opinion Monitor Artificial Intelligence].
We thank the Observatory for Artificial Intelligence in Work and Society, a project of the Policy Lab Digital Work & Society, at the
Federal Ministry of Labor and Social Affairs, for their cooperation in developing questions for the special survey.
Meinungsmonitor Künstliche Intelligenz
Opinion Monitor Artificial Intelligence
Contact person: Pero Dosenovic, M. A.
Email: memoki@hhu.de
Website: cais.nrw/memoki
Project by: Funded by:You can also read
